Communications inspired linear discriminant analysis

We study the problem of supervised linear dimensionality reduction, taking an information-theoretic viewpoint. The linear projection matrix is designed by maximizing the mutual information between the projected signal and the class label. By harnessing a recent theoretical result on the gradient of mutual information, the above optimization problem can be solved directly using gradient descent, without requiring simplification of the objective function. Theoretical analysis and empirical comparison are made between the proposed method and two closely related methods, and comparisons are also made with a method in which Rényi entropy is used to define the mutual information (in this case the gradient may be computed simply, under a special parameter setting). Relative to these alternative approaches, the proposed method achieves promising results on real datasets. Copyright 2012 by the author(s)/owner(s).

Real-Time and Data-Driven Operation Optimization and Knowledge Discovery for an Enterprise Information System

An enterprise information system (EIS) is an integrated data-applications platform characterized by diverse, heterogeneous, and distributed data sources. For many enterprises, a number of business processes still depend heavily on static rule-based methods and extensive human expertise. Enterprises are faced with the need for optimizing operation scheduling, improving resource utilization, discovering useful knowledge, and making data-driven decisions.

This thesis research is focused on real-time optimization and knowledge discovery that addresses workflow optimization, resource allocation, as well as data-driven predictions of process-execution times, order fulfillment, and enterprise service-level performance. In contrast to prior work on data analytics techniques for enterprise performance optimization, the emphasis here is on realizing scalable and real-time enterprise intelligence based on a combination of heterogeneous system simulation, combinatorial optimization, machine-learning algorithms, and statistical methods.

On-demand digital-print service is a representative enterprise requiring a powerful EIS.We use real-life data from Reischling Press, Inc. (RPI), a digit-print-service provider (PSP), to evaluate our optimization algorithms.

In order to handle the increase in volume and diversity of demands, we first present a high-performance, scalable, and real-time production scheduling algorithm for production automation based on an incremental genetic algorithm (IGA). The objective of this algorithm is to optimize the order dispatching sequence and balance resource utilization. Compared to prior work, this solution is scalable for a high volume of orders and it provides fast scheduling solutions for orders that require complex fulfillment procedures. Experimental results highlight its potential benefit in reducing production inefficiencies and enhancing the productivity of an enterprise.

We next discuss analysis and prediction of different attributes involved in hierarchical components of an enterprise. We start from a study of the fundamental processes related to real-time prediction. Our process-execution time and process status prediction models integrate statistical methods with machine-learning algorithms. In addition to improved prediction accuracy compared to stand-alone machine-learning algorithms, it also performs a probabilistic estimation of the predicted status. An order generally consists of multiple series and parallel processes. We next introduce an order-fulfillment prediction model that combines advantages of multiple classification models by incorporating flexible decision-integration mechanisms. Experimental results show that adopting due dates recommended by the model can significantly reduce enterprise late-delivery ratio. Finally, we investigate service-level attributes that reflect the overall performance of an enterprise. We analyze and decompose time-series data into different components according to their hierarchical periodic nature, perform correlation analysis,

and develop univariate prediction models for each component as well as multivariate models for correlated components. Predictions for the original time series are aggregated from the predictions of its components. In addition to a significant increase in mid-term prediction accuracy, this distributed modeling strategy also improves short-term time-series prediction accuracy.

In summary, this thesis research has led to a set of characterization, optimization, and prediction tools for an EIS to derive insightful knowledge from data and use them as guidance for production management. It is expected to provide solutions for enterprises to increase reconfigurability, accomplish more automated procedures, and obtain data-driven recommendations or effective decisions.

CT Scan Dosimetric Parameters Routine Monitoring: First Results of Radiation Dose Optimization Strategies Promptly Provided by a Multidisciplinary Team

CONCLUSION Radiation dose reduction, while saving image quality could be easily implemented with this approach. Furthermore, the availability of a dosimetric data archive provides immediate feedbacks, related to the implemented optimization strategies. Background JCI Standards and European Legislation (EURATOM 59/2013) require the implementation of patient radiation protection programs in diagnostic radiology. Aim of this study is to demonstrate the possibility to reduce patients radiation exposure without decreasing image quality, through a multidisciplinary team (MT), which analyzes dosimetric data of diagnostic examinations. Evaluation Data from CT examinations performed with two different scanners (Siemens DefinitionTM and GE LightSpeed UltraTM) between November and December 2013 are considered. CT scanners are configured to automatically send images to DoseWatch© software, which is able to store output parameters (e.g. kVp, mAs, pitch ) and exposure data (e.g. CTDIvol, DLP, SSDE). Data are analyzed and discussed by a MT composed by Medical Physicists and Radiologists, to identify protocols which show critical dosimetric values, then suggest possible improvement actions to be implemented. Furthermore, the large amount of data available allows to monitor diagnostic protocols currently in use and to identify different statistic populations for each of them. Discussion We identified critical values of average CTDIvol for head and facial bones examinations (respectively 61.8 mGy, 151 scans; 61.6 mGy, 72 scans), performed with the GE LightSpeed CTTM. Statistic analysis allowed us to identify the presence of two different populations for head scan, one of which was only 10% of the total number of scans and corresponded to lower exposure values. The MT adopted this protocol as standard. Moreover, the constant output parameters monitoring allowed us to identify unusual values in facial bones exams, due to changes during maintenance service, which the team promptly suggested to correct. This resulted in a substantial dose saving in CTDIvol average values of approximately 15% and 50% for head and facial bones exams, respectively. Diagnostic image quality was deemed suitable for clinical use by radiologists.

Acoustic and Magnetic Techniques for the Isolation and Analysis of Cells in Microfluidic Platforms

Cancer comprises a collection of diseases, all of which begin with abnormal tissue growth from various stimuli, including (but not limited to): heredity, genetic mutation, exposure to harmful substances, radiation as well as poor dieting and lack of exercise. The early detection of cancer is vital to providing life-saving, therapeutic intervention. However, current methods for detection (e.g., tissue biopsy, endoscopy and medical imaging) often suffer from low patient compliance and an elevated risk of complications in elderly patients. As such, many are looking to “liquid biopsies” for clues into presence and status of cancer due to its minimal invasiveness and ability to provide rich information about the native tumor. In such liquid biopsies, peripheral blood is drawn from patients and is screened for key biomarkers, chiefly circulating tumor cells (CTCs). Capturing, enumerating and analyzing the genetic and metabolomic characteristics of these CTCs may hold the key for guiding doctors to better understand the source of cancer at an earlier stage for more efficacious disease management.

The isolation of CTCs from whole blood, however, remains a significant challenge due to their (i) low abundance, (ii) lack of a universal surface marker and (iii) epithelial-mesenchymal transition that down-regulates common surface markers (e.g., EpCAM), reducing their likelihood of detection via positive selection assays. These factors potentiate the need for an improved cell isolation strategy that can collect CTCs via both positive and negative selection modalities as to avoid the reliance on a single marker, or set of markers, for more accurate enumeration and diagnosis.

The technologies proposed herein offer a unique set of strategies to focus, sort and template cells in three independent microfluidic modules. The first module exploits ultrasonic standing waves and a class of elastomeric particles for the rapid and discriminate sequestration of cells. This type of cell handling holds promise not only in sorting, but also in the isolation of soluble markers from biofluids. The second module contains components to focus (i.e., arrange) cells via forces from acoustic standing waves and separate cells in a high throughput fashion via free-flow magnetophoresis. The third module uses a printed array of micromagnets to capture magnetically labeled cells into well-defined compartments, enabling on-chip staining and single cell analysis. These technologies can operate in standalone formats, or can be adapted to operate with established analytical technologies, such as flow cytometry. A key advantage of these innovations is their ability to process erythrocyte-lysed blood in a rapid (and thus high throughput) fashion. They can process fluids at a variety of concentrations and flow rates, target cells with various immunophenotypes and sort cells via positive (and potentially negative) selection. These technologies are chip-based, fabricated using standard clean room equipment, towards a disposable clinical tool. With further optimization in design and performance, these technologies might aid in the early detection, and potentially treatment, of cancer and various other physical ailments.

Bayesian Emulation for Sequential Modeling, Inference and Decision Analysis

The advances in three related areas of state-space modeling, sequential Bayesian learning, and decision analysis are addressed, with the statistical challenges of scalability and associated dynamic sparsity. The key theme that ties the three areas is Bayesian model emulation: solving challenging analysis/computational problems using creative model emulators. This idea defines theoretical and applied advances in non-linear, non-Gaussian state-space modeling, dynamic sparsity, decision analysis and statistical computation, across linked contexts of multivariate time series and dynamic networks studies. Examples and applications in financial time series and portfolio analysis, macroeconomics and internet studies from computational advertising demonstrate the utility of the core methodological innovations.

Chapter 1 summarizes the three areas/problems and the key idea of emulating in those areas. Chapter 2 discusses the sequential analysis of latent threshold models with use of emulating models that allows for analytical filtering to enhance the efficiency of posterior sampling. Chapter 3 examines the emulator model in decision analysis, or the synthetic model, that is equivalent to the loss function in the original minimization problem, and shows its performance in the context of sequential portfolio optimization. Chapter 4 describes the method for modeling the steaming data of counts observed on a large network that relies on emulating the whole, dependent network model by independent, conjugate sub-models customized to each set of flow. Chapter 5 reviews those advances and makes the concluding remarks.

Perioperative fluid therapy: a statement from the international Fluid Optimization Group.

BACKGROUND: Perioperative fluid therapy remains a highly debated topic. Its purpose is to maintain or restore effective circulating blood volume during the immediate perioperative period. Maintaining effective circulating blood volume and pressure are key components of assuring adequate organ perfusion while avoiding the risks associated with either organ hypo- or hyperperfusion. Relative to perioperative fluid therapy, three inescapable conclusions exist: overhydration is bad, underhydration is bad, and what we assume about the fluid status of our patients may be incorrect. There is wide variability of practice, both between individuals and institutions. The aims of this paper are to clearly define the risks and benefits of fluid choices within the perioperative space, to describe current evidence-based methodologies for their administration, and ultimately to reduce the variability with which perioperative fluids are administered. METHODS: Based on the abovementioned acknowledgements, a group of 72 researchers, well known within the field of fluid resuscitation, were invited, via email, to attend a meeting that was held in Chicago in 2011 to discuss perioperative fluid therapy. From the 72 invitees, 14 researchers representing 7 countries attended, and thus, the international Fluid Optimization Group (FOG) came into existence. These researches, working collaboratively, have reviewed the data from 162 different fluid resuscitation papers including both operative and intensive care unit populations. This manuscript is the result of 3 years of evidence-based, discussions, analysis, and synthesis of the currently known risks and benefits of individual fluids and the best methods for administering them. RESULTS: The results of this review paper provide an overview of the components of an effective perioperative fluid administration plan and address both the physiologic principles and outcomes of fluid administration. CONCLUSIONS: We recommend that both perioperative fluid choice and therapy be individualized. Patients should receive fluid therapy guided by predefined physiologic targets. Specifically, fluids should be administered when patients require augmentation of their perfusion and are also volume responsive. This paper provides a general approach to fluid therapy and practical recommendations.